Episode 8: AI Readiness for Salesforce Podcast

“The good part is that actors like OpenAI really evangelize the market and democratize the usage of solutions like ChatGPT. So people are now trained to use that kind of tool within their day to day activities. But here come the bad things. The usage of those technologies does not follow governance rules or security concerns. So what you you are doing with your individual use cases at home, leveraging AI and ChatGPT, cannot be just copy and pasted into your company, because you need to take care about your data, the data of your customer, your processes, the governance, the regulations.” – Cyril Louis, Mavericx

We were delighted to host new Hall of Fame Salesforce MVP, Cyril Louis, CEO & Co-Founder at Salesforce consultancy Mavericx in Episode 8 of our AI Readiness for Salesforce Podcast.

Video transcript

Hello, everybody. Hello. Hello and welcome to another episode of our AI readiness for Salesforce. It’s a quite exciting one today. We have, Cyril, with us, you know, one of the top experts, definitely in Europe, probably in the world, whatever is going in the Salesforce ecosystem, specifically relating to AI and Agentforce and the evolution of that conversation.

So, Cyril, thank you very much for joining us. Would you care introducing yourself for our audience?

Yeah, sure. Well, first of all, thank you for that opportunity. That invitation. So, yeah, my name is Cyril. I am, working in this ecosystem, Salesforce ecosystem since 15 plus years now. So it’s been a long journey.

I’ve seen a lot of things. I am still enjoying the platform and learning on the platform. And within that journey, I’m also sharing a lot with the community. So being a user group leader here in Geneva, Switzerland, being recognized as the Salesforce MVP. I often speak out in different conferences. And more recently, also the founder of Mavericx, an implementation partner consulting firm to guide the customer to their Salesforce implementation to achieve the success.

Well that’s fantastic. Yes. And I think, you know, there was a recent milestone in your effort to spread the word and your wisdom. And, you know, what you learn about the ecosystem and package it so that everybody can access to it through your LinkedIn. I believe you celebrated recently a major milestone, right?

Yeah. I’m old enough on the whole ecosystem to to now reach the Salesforce MVP Hall of Fame.

So, the Hall of Fame. Fantastic. And then 10,000 followers on LinkedIn.

Yeah. In there also. Yeah. Another milestone. I was not checking that often. I just realized and said, oh, I forgot. I moved on that platform. I used to work X, formerly known as Twitter, but I’m not publishing and sharing a lot in there now.

Understood. And so. Well, your LinkedIn is, is a great source of information for me. And I’m sure for a lot of people, 10,000 of us, at least. So I invite everybody to, you know, check out, LinkedIn. We’ll we’ll actually link it in the, text, that goes with, with this episode. Now, Cyril… AI readiness, a big topic, right? And, just earlier we were talking about AI readiness, and you mentioned something that would be probably be a very good start to our conversation. So you said, “When I talk to companies and when I talk to the broader ecosystem in Salesforce, I often mentioned that being ready for AI is not only related to technology.” So can you care to elaborate on that? And, you know, let’s start with that.

Yeah, yeah, I told them that, AI is not just about technology. I would say it’s something like 30% technology, 30% processes, and 40% the people. You need to understand the process that you would like to automate or to leverage with AI, but then you need to train your people. You need to raise confidence on what will be the AI prediction or recommendation.

If they don’t trust a platform, they will not care about AI. And this is also about relying the data on which the AI will be based. So again, it’s not just about technology. It’s a mix and too often companies, maybe because of FOMO, they will like to jump into a solution and they see AI as a product, not as a process. And they just said, oh, I would like to to try this. You know, I buy it, I activate and then, let the magic happen. It’s not that easy. Unfortunately.

It’s interesting you put, you know, the majority of the components is in the people, which is which is interesting. I read, a stat recently that research that is saying that 80% of the people at work, they already use AI. And it’s an interesting stat because it triggers a number of observations, right? So and I think they’re mainly referring probably to using GPT or using an AI notetaker in their calls or, you know, some. And what are the implications and is that, do you think, good or bad for building an AI strategy?

It is exactly good *and* bad.

The good part is that actors like OpenAI really evangelize the market and democratize the usage of solutions like ChatGPT. So people are now trained to use that kind of tool within their day to day activities. But here come the bad thing. The usage of those technology does not follow governance rules or security concerns. So what you you are doing with your I would say individual use cases at home, leveraging AI and ChatGPT, cannot be just copy and paste into your company because you need to take care about your data, the data of your customer, your processes, the governance, the regulations.

So we have a set of additional layers on top of this. And this is the thing some of the users would like to just use it like this with not taking care about governance or security. So this is the bad thing.

So can I be a bit of a devil’s advocate here because then recently in my interactions with, you know, I can think of the, a couple of companies, actually one German and one Belgian. They find themselves, they believe exactly in that limbo stage where on the one hand, they dread looking at their team and everybody’s using their own AI and a lot of free tools and probably by doing so, leaking a lot of information on the should be under control. And it’s not so that they are trying to put a stop to it, you know, knowing that they’re stopping possibly productivity on the one hand. On the other hand, they’re not ready to embrace AI or to put governance around the AI because they are lost in committees and conversations and regulations and the UAI act and all that comes with it. And so this is, I believe, a limit for European companies, because especially compared to US companies that seem to be accelerating through through AI. Not caring too much and possibly exposing them maybe themselves to a bigger risk and not caring too much about governance. So which one is which? And, you know, how do we help solve this? Or, you know, let me start with, do you think my analysis is is is correct, at least in some cases, you know.

Yeah. So yeah, we indeed have, especially in Europe, additional compliance to… First is regarding the customer data, but also regarding I would say AI usage, ethics and cybersecurity. So we need to have this kind of ethical guardrails on top of the usage and cybersecurity. So how do you secure the data that you will be using. And and the one that you will be sending to an external system to feed the AI model? This is definitely an additional layer of complexity for some of the companies. And and this is also the switch or the, the difference between the mindset of two types of companies. We have the big corporates that, they don’t they have no choice. They need to follow governance and compliance and security review. So they will lose velocity.

And we have smaller company that they can have like a quick start approach, more use cases, focus to see the value. Maybe I would say more pragmatic approach. Let’s start with something. And then we will iterate to grow our usage before to to start something too big. So yeah, not sure which which, which strategy is the best. It’s that we need sometimes you to follow global regulations. So you know of course, of course. And I, and I guess, yeah, they, they have to. Right.

So as you said, you know, the they are part of a regulated environment and they cannot NOT take care of it. And that’s probably where Salesforce can play and does play a major role, because they obviously bring a very structured approach to their value proposition around AI with Agentforce, the trust layer, the controls and guardrails around the implementation elements of, you know, how data is handled.

You have examples of actually that being a successful deployment and, you know, kind of where where companies on a step by step approach, I guess, in Europe are starting to embrace this type of AI within the Salesforce example.

Yes. And then actually it was a, a journey, not just a one off project saying I would like to go with AI on Salesforce. And this is probably this is why it was a success,

They started with the foundation regarding where are the data that I will need? Do I have access to that data? Is that data relevant? How is synchronize the data that is coming from an external system? Is it synchronized often enough? Then how do you merge all those different data sources to have a clear understanding of the global customer? And then when everything has been cleaned up or harmonized, how to activate AI on top of this? Because actually AI is a model that will need a context. And the context come from historical data. Yeah. So first you need to be to be clean on your data. And then you can leverage AI. So the success that I’ve seen, it was this kind of journey capturing the data, harmonizing the data and then activating AI use cases and I agree with you regarding how Salesforce can be an accelerator with that kind of use cases, because Salesforce itself thought about that project as an iterative process.

You can activate Agentforce with data only on your CRM. It’s a kind of, straightforward process. Then you can increase the knowledge with additional context coming from external sources and structured data. So for that you will need more on top of Agentforce. You will need a bit of Data Cloud.

And then you can increase the way to interact with your data with conversional agents, not just AI process and automated.

So you can activate step by step and you will mature your process and you will your AI readiness, in order to make it to success.

I very much like the, the concept of being, of it being a journey. I’m actually curious if it’s the right thing. Right. So we’ve seen it across many use cases and many, demos as well and success stories that we’ve seen on stage at the recent events with, with Salesforce, that the foundation is paramount and it’s all about data, but it takes it takes a strategic mind and a long term approach to actually start, with that.

And, so I’m wondering, in your experience, are client clients that strategic when they think about Agentforce and the future of their business and the, you know, kind of starting with data, or is it something that you have to bring as part of your expertise and kind of work with them to? Because I would imagine that as a client, I’m thinking, well, I need to ChatGPT for my team that does this, and they start at the very end. And so is that your job? You know, kind of reeling it back to the foundation layer?

I will start with a pure consulting answer. It depends.

So I have I’ve seen different customers, obviously. It depends their maturity and their vision. Unfortunately, the vast majority, I would say, have the wrong way to address that process.

And it might be because Salesforce, in their promotion of the product, try to evangelize a bit too much and make it more a product that you just act-, you buy it and then you activate it and it’s done. So this is somehow the mindset that I’ve seen too much is not the correct mindset, the quick win or the proof of concept approach. Let’s start with this as soon as possible, and then we will see where we land.

On another hand, we have customers with true maturity, a clear vision and they say we want to go at scale. So it has to be security by design, but everything has to be documented. The way our agent will, will react, is it exactly the behavior that I would like him to have? Where are my data? What is the context? Everything had to be thought before to implement. So we have a lot of duties to be, to, to be done before the implementation. And I’ve seen too many customers jumping on, activating something and just maybe hoping that it will be part of the magic that Salesforce is showcasing during all these amazing conference.

They try to sometimes oversimplify, the behind the scene. Yeah, I think we we’ve we’ve learned that the hard way sometimes actually throughout the years, right? With the, the beautiful demos and then the client saying “I want that” and I’m like, okay, let me take you on the journey that that totally makes sense.

Salesforce very often, uses, a very interesting concept of customer zero. So they often refer to themselves as, the first customer. So they build solutions. So they, they first in the first place, they can use that you know in their sales and marketing organization. So it makes a lot of sense to test on themselves. Do you do does Mavericx, also kind of take a customer zero approach? Have you started looking at implementing some of the AI capabilities across your own organization?

Yes. Clearly for for many reasons. The first one is the the customer should be a bit of, innovative. Let’s say let’s use the, the Salesforce lexi, a trailblazer, you know, open to the innovation. So this is the first mindset that we are looking for because it’s it’s a bit of a bet on something that has not been implemented yet.

So first they need to align with, the vision of Salesforce. Yes. I’m confident enough that we can reach what Salesforce is, is demoing is showing that we can land there. So it’s the first criteria. The second one is the trust with the implementation partners. It’s a it’s a new product. It’s a new mindset. It’s a new way of working.

Probably we have the knowledge of the the global solution. And we have something new coming up.

So they need to trust us that we will learn together. It will be iterative. So it’s as you said, the zero approach, the zero customer, the one that trust us, aligned with, with with the Salesforce vision and that trust us to start something and then we will see where we can go.

And most of the case initial vision is not exactly aligned with, the operation, the, the way to, you know, the vision to operation. We, we have a gap because the product sometimes is not mature enough, sometimes because our own expertise is not at the level needed to fine tune the product itself. And sometimes it’s a bit of the those two criteria.

And then as soon as we have a first use cases in production, first we have the return of experience. We touch, we play with the product. We know accelerators, best practices, mistake not to to do again and again. And it’s an accelerator for additional customer that the the the, the product the the the selling will be easier.

But you’re right, the first one will be someone that has the product that has a vendor and that has the implementation for something that I’ve never been implementing before. it’s an interesting area and goes back to the, you know, you need to to choose and select, a journey partner, right? So the partner that we sit, you know, with you alongside you and drive, with you, the project forward.

Look, I wanted to jump, into another interesting area. So following you on, on LinkedIn, and thinking about impact of a recent announcement, recent evolutions I wanted to pick your brain on, on a few, you know, quick snapshot areas. One is, regarding, the change of pricing, you know, big announcement, arriving, you know, you know, what do you think about it? Another one is the Informatica acquisition and, you know, major, major, player, in the market, you in your narrative around that acquisition, you you focus a lot again on data on the foundation, of, you know, building, building AI success. And then there was another, very interesting, piece of, of, information that you put out there which, analyze the difference, which I think is still not clear. I think it’s part of your education to the market. That looks at the differences between, the old AI, kind of conversational AI, a copilot and an agent, and kind of the evolution of that concept and the impact expected.

So I’m, I want to pick your brain and, you know, kind of a flash on all of these. And then we look into the future for a sec.

So I will start with. Yeah, we’ll start with the Informatica, acquisition. I like that one because it’s something that I’ve shared a year ago. So surprisingly, a year ago, and I was mentioning it would have been a good bet for Salesforce.

And I explained why it was not a duplicate buying Informatica on top of MuleSoft. And I and I explained in, when I was sharing my thoughts that Informatica is much more. It can be seen as an MDM, and one of the the key differentiator is the the transformation capabilities. When we come back to the global suite of Salesforce, I would say two major missing pieces, especially in Data Cloud was data transformation.

Today, the product is not as the level of excellence that Salesforce used to, to deliver, when you want to do data transformation in Data Cloud, it is a mess. The product is not really meant for that. It’s a lot of workarounds. So that could be a great add-on because Informatica is doing this perfectly. And the second one is the MDM approach.

A lot of customers and sometimes even Account Executive think Data Cloud is meant to have this unified customer. The correct wording for Data Cloud is much more a kitchen. So you you will have different records grouped together to have a better understanding of your customer. But all all those records that we kind of match thanks to identity resolution, they are not merged.

So this is why we said the Data Cloud is not a unified customer. It’s a unified view. Yes, it’s better understanding. You can calculate information on top of this because you have a better knowledge and understanding. But Informatica, on the other hand is a solution for MDM. And that will be the master data of your solution that will merge the data.

So here again could be an additional asset on top of Data Cloud or to complete the solution of Data Cloud.

Very very interesting. And again, you know, data kind of coming back in, you know, it’s a very topical for, for us always, you know, doing our episodes about AI readiness because I think, that’s where, the, the battle is, is, is played.

For sure then. Yeah. I always, used to say data is as good as your, No AI is as good as your data. So, obviously, data is key. The foundation on which AI will be will be built. Yeah. For sure.

Pricing. What do you think? Actions, conversations are not priced if there’s no action, but actions are price individually $0.10 credits. Flexible credits. All good?

Yeah, yeah, yeah. I will start this topic on a with another point of view. It’s having something that you pay by seat compare by usage is the first mindset switch that we see on the market… And is the customer, is the market really for that? And the part of the answer will be again, it depends because depends to who you’re talking to.

The department that you are interacting with for marketing departments, actually, they are already used to have this kind of usage invoicing because they are per volume. How many emails you are sending, how many segment you using, how many SMSs you’re sending. So they are already used to that usage invoicing. For IT or for sales they are coming from a world where you buy a solution, and then you use it as much as you can to increase the return on investment. And so for for those departments, the switch is complex. So it’s just a maturity. So of course some the vendors their role is to educate and to show the value. And this is also why currently we are seeing Salesforce and other vendors trying to initiate new pricing model to see how the market will react to that pricing model.

Is it is the correct one or should we change something and is exactly why Salesforce is reactive enough to switch and to change their pricing model. So the market might say it’s not clear. It’s not accurate. It’s not stable enough Salesforce is changing again and again, but actually just to react accordingly to the expectation of the market Conversation paying per conversation is not meaningful because sometimes the conversation is out of context. We don’t create anything out of out of this conversation. So actions is more useful for as a KPI. But then again it’s is your agent used internally? So user agent or exposed externally customer agent and here again you can see different ways to price your agents and how many data are used to to create the result of the agent.

Are they coming from Salesforce or are they coming from elsewhere? And then you are a complexifying the pricing model. At the beginning, it was to make it more relevant, but if you want to be more relevant, you are adding criteria in the pricing model. If you are adding criteria, you are increasing the complexity of the understanding of the pricing model. So this is exactly what we are. It sounds it’s more legit to go that way, but if it’s even more complex for a customer to understand the pricing model, we said to the last conversation, super easy, but now we want to mean more relevance. So only price when I see a value. So you add criteria is more more complex to understand.

Yeah, true. I think the more we get educated and educate ourselves, you know, in the market and the more likely we will evolve towards an outcome-based pricing, right? So you started to hint to it when because the action is closer to an impact than when you when you just focus on the impact it will be, I think the Holy Grail.

Right. So you know exactly what you pay for because you have an outcome. Unless there’s an outcome, there’s no cost to me, which I think it’s a nirvana state for for any pricing for any company. And Salesforce now is also introducing the flex credit, as you said. So it’s another way to align with the market expectations. We all start with assumptions.

So what if we are not on that vision? So the flex credit is just a way to estimate your consumption. But then if you’re not on it, the flex credit can be reused later on on another use cases. And then Salesforce bring you it kind of dashboards to control how to use those credits. Is it for that use case, that use case, that use case? So you can also orchestrate where you would like to allocate more flex credits on the use cases, or the processes where you see more value. And here again, for the customer, I don’t want my “forget password agents” to consume all of my flex credit. Not meaningful for me, so you can reallocate this.

And now you have a dashboard to allocate it, but also track and deep dive into all the records that have been used by your credit. Salesforce is giving you more visibility, more transparency on how your you can use and where your credits are used.

Well, sounds, very, very interesting and promising, actually. And, I think we are all, rooting from, from the sideline right? So actually, last, you know, 30s, you know, we’ve taken a lot of your knowledge already and time. So I really appreciate it. But, looking into the future, are you, you know, bullish are you skeptical? What do you see? What’s what does your, you know, kind of, you know, sphere glass sphere what’s the crystal, you know, ball say.

Yeah.

I am not selling the technology. I’ll say I’m much more finding use cases to use the technology to create value, productivity to the customers. So I am not involved into the pricing model. And in fact, of course, if the market don’t find any value, they will not go for it. And I will have nothing to implement.

But on the implementation part of you, I to take your wording, I’m bullish and I’m sure that the usage will increase, and especially because the technology is evolving very fast. And I think we will go more and more on this low code, no code way to implement the technology as a facilitator.

The only thing is because you can do it it’s not because you can that you should, you know? So it’s you will be able to implement a lot of things with no technology, technology background. The only thing is how to document this. How do you make sure that this will scale? How do you anticipate the impact of one APEX class or one use case on the way your agent will behave.

So this is that the process and as I told you, is 30% technology, 30% process and 40%, the people. So don’t forget those two, those two, pillars.

And I think more and more your role role and your job and your team’s job will be to safeguard those aspects, right? And bring your expertise in those areas. Educate and guide.

Yes, I agree with you. You know, wow. This was, very, very good conversation. Thank you very much Cyril, as always, a lot of learning, a lot of, very current, very practical, very impactful information, from you. Thank you very much. Thank you. And, I will I will see you soon.